[21pt] Quantify WSE differences at max ras2fim discharge

[CarsonPruitt-NOAA]

Use the maximum discharge from the exported ras2fim rating curves to feed into the HAND rating curves and compare the two WSE. We hope to see an insignificant difference between the two, but this evaluation will determine if we need to take another look at the calibration technique.

[AliForghani-NOAA]

Below is an analysis on the Water Surface Elevation (WSE) comparisons between ras2fim VS HAND run for 12090301. The analysis is performed on the rating curve points derived from ras2fim run (_reformat_ras_rating_curvepoints.gpkg file). Below we see a map of the ras models points on the DEM from HAND run:

The ras WSE is directly obtained from ras rating curves file (_reformat_ras_rating_curvetable-12090301.csv) for the maximum discharge value. HAND WSE is computed by summing the two items below:

• DEM datum that is sampled for ras points on the DEM
• Interpolated Stage estimated from _hydroTable0.csv using the given max discharge for the HydroID containing the ras points

Below we see some plots comparing the two methods to estimate WSE. Roughly speaking, HAND WSEs are 1.8m higher than ras WSE (RMSE=1.8m):

The "error" --difference between ras and HAND-- is higher for low WSE values:

If we use the 'precalb_discharge_cms' instead of 'discharge_cms' the plots change to

[AliForghani-NOAA]

New Analysis

The idea is to update the HAND SRC using ras2fim results as explained below:

• For the minimum flow from ras2fim rating curve, find minimum WSE from HAND SRC.
• Compare the minimum WSE of HAND with ras2fim, and call the difference 'WSE_base_error". (for example, 5m in plot below)
• Convert WSE results of ras2fim rating curve into stage (by subtracting DEM from WSE)
• Use ra2fim discharge-stage and look-up the discharge corresponding to "WSE_base_error", and call it "Q_Adjust"
• Add the "Q_Adjust" flow to HAND SRC

[CarsonPruitt-NOAA]

Instead of converting the ras2fim rating curve to stage, you should:

• Take the adjusted DEM value from HAND--assume HAND minimum stage to be zero.
• Interpolate a discharge from the ras2fim rating using the adjusted DEM elevation (aka HAND minimum WSE)
• Add that discharge to all of the flows in the HAND rating curve

Note: Apply above methodology only when HAND minimum WSE is higher than ras2fim minimum WSE.

[AliForghani-NOAA]

Revised Methodology

The goal is to update HAND SRC using ras2fim points results as explained below:

• For ras2fim points, HAND minimum WSE is assumed as the adjusted HAND DEM
• Find HAND adjustment discharge for ras2fim points
  • Method 1: Interpolate a discharge from the ras2fim rating curves (flow vs WSE) for the HAND minimum WSE (aka adjusted DEM elevation), and call it "Q_Adjust"
  • Method2
    • Compare the minimum WSE of ras2fim with HAND, and call the difference 'WSE_base_error".
    • Add stage values into ras2fim rating curves (by subtracting DEM from WSE)
    • Interpolate a discharge from ras2fim discharge-stage rating curves for the "WSE_base_error", and call it "Q_Adjust"
• Compute the median of "Q_Adjust" of ras2fim points within each HAND HydroID
• Add the "Q_Adjust" median value into flows of HAND SRC

Evaluation

The above methods to update HAND SRC are evaluated using the max flow of ras2fim rating curves. For each ras2fim point, “Error” is computed as the difference between maximum WSE from ras2fim rating curve and HANDs interpolated WSE for max flow of ras2fim. The adjustment methods are applied only when original HAND max WSE is higher than ras2fim max WSE.

• Original Error = RAS_max_WSE minus Original HAND max WSE
• Method1_Error = RAS_max_WSE minus HAND max WSE after Method 1 Adjustment
• Method2_Error = RAS_max_WSE minus HAND max WSE after Method 2 Adjustment

Error reduction is computed using Equations below:

• Method1_Error_Reduction=100*("Orig_Error"-"Method1_Error")/"Orig_Error"
• Method2_Error_Reduction =100*("Orig_Error"-"Method2_Error")/"Orig_Error"

On ras2fim points level, as we showed earlier, RMSE of the original error is 1.803m. The error is reduced to 1.789m and 1.756m for Method1 and Method2, respectively. This is 0.8% and 2.64% error reduction for Methods 1 and 2, respectively.

Also, we can aggregate the errors of ras2fim points for each HAND HydroID by applying RMSE equation. Plot below compares error reduction percentages between the two methods: